LSTM ネットワークを用いたロボットの移動経路の学習と探索

抄録

We propose a novel robot path planning method combining RRT and LSTM network. In this method, numerous and good paths are generated in the robot configuration space by RRT method, a small size LSTM network is trained by them. By the proposed method, the difficulty of general methods with LSTM network, i.e. “the acquisition of a large amount of training data” is overcame. Moreover, the difficulty of general random based methods, i.e. “the reproducible path generation” is enabled with high-speed.